Purpose This paper aims to address the practical realization gap that arises when second-order position-based visual servoing (SO-PBVS), designed at the acceleration level, is executed on commercial industrial manipulators through sampled Cartesian-pose commands. Design/methodology/approach A realization-aware bounded-funnel-inspired SO-PBVS method is developed for pose-command manipulators. The nominal SO-PBVS law is embedded in a reference-twist layer with damping leakage, saturation-consistent anti-windup and discrete SE(3) pose updates. Bounded funnel-shaped and terminal terms are used as saturated performance-shaping corrections. The method is evaluated on an RM65-B eye-in-hand platform using AprilTag pose feedback and repeated hardware trials. Findings Under a severe initial error of about 0.216 m and 0.667 rad, directly realized SO-PBVS did not maintain the strict final accuracy. Realization-aware performance-funnel-SO-PBVS entered the strict accuracy band in all repeated trials. Compared with RA-Damped SO-PBVS, it reduced strict first-entry time from 17.55 s to 15.51 s, reduced final-window rotational error from 0.0175 rad to 0.0095 rad and increased strict-band retention from 67.3% to 96.7%. Originality/value The work shows that command-realization effects can dominate SO-PBVS final-stage behavior on pose-command industrial manipulators and provides an interface-compatible realization layer and bounded correction structure for practical visual-servoing implementation.
No takes yet. Share an insight, caveat, or question.
Gao et al. (2026) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: